toksuite / test_checkpoint /train_state_00001.json
craffel's picture
craffel HF Staff
Upload folder using huggingface_hub
7c6481d verified
Raw History Blame Contribute Delete
2.22 kB
{"step": 50000, "acc_step": 0, "data_loader_state": {"it_state": {"start_token": 272, "it_state": {"it_state": {"root_dir": "/scratch/craffel/lingua/data/tokenizer_training/", "sources": {"fw_edu": 0.5, "cmn_Hani": 0.1, "tur_Latn": 0.1, "ita_Latn": 0.1, "fas_Arab": 0.1, "stack_edu": 0.1}, "source_to_state": {"fw_edu": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/fw_edu/fineweb_edu_100bt.chunk.01.jsonl", "position": 18011435196, "block_size": 1, "offset": 0, "current_iter": 0}, "cmn_Hani": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/cmn_Hani/fineweb_2_hq.cmn_Hani.chunk.01.jsonl", "position": 2982960015, "block_size": 1, "offset": 0, "current_iter": 0}, "tur_Latn": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/tur_Latn/fineweb_2_hq.tur_Latn.chunk.01.jsonl", "position": 2437300342, "block_size": 1, "offset": 0, "current_iter": 0}, "ita_Latn": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/ita_Latn/fineweb_2_hq.ita_Latn.chunk.01.jsonl", "position": 2603201448, "block_size": 1, "offset": 0, "current_iter": 0}, "fas_Arab": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/fas_Arab/fineweb_2_hq.fas_Arab.chunk.01.jsonl", "position": 331901344, "block_size": 1, "offset": 0, "current_iter": 1}, "stack_edu": {"file_path": "/scratch/craffel/lingua/data/tokenizer_training/stack_edu/stack_edu.chunk.01.jsonl", "position": 3676886775, "block_size": 1, "offset": 0, "current_iter": 0}}, "rng_state": {"bit_generator": "PCG64", "state": {"state": 75499279334076499819138845636886208810, "inc": 246509925186285949978196491240064802315}, "has_uint32": 0, "uinteger": 0}}, "add_bos": true, "add_eos": true, "name": "huggingface", "path": "google/gemma-2-2b"}, "output_seq_len": 4096, "n_views": 2}, "seq_idx": 640, "rng_state": {"bit_generator": "PCG64", "state": {"state": 129376025135317909054550778266736994970, "inc": 173555323965545256606922338259303677603}, "has_uint32": 1, "uinteger": 2164706399}, "batch_size": 4, "prefetch_size": 1024}, "scheduler": {"base_lrs": [0.001], "last_epoch": 50000, "verbose": false, "_step_count": 50001, "_get_lr_called_within_step": false, "_last_lr": [7.157234034479355e-05], "lr_lambdas": [{}]}}